update-skills
Skill13 starsChanged 44 days agoMITView on GitHub
Check vendored skills for upstream updates and selectively apply requested updates while preserving local adaptations and provenance. Use when the user asks whether third-party skills are outdated, wants an upstream comparison, or asks to refresh installed skill copies. For directory structure and catalogue maintenance alone, use update-skills-directory.
20 other projects use a version of this.
Adopted by
Most recently taken up by
- antick/skillstheir own version, first seen Nov 16, 2026
- Trixx4191/asrartheir own version, first seen Nov 15, 2026
- nanocoai/nanoclawtheir own version, first seen Nov 15, 2026
- IndusCore-AI/connect-anythingtheir own version, first seen Nov 8, 2026
- tandregbg/core-claude-skillstheir own version, first seen Nov 4, 2026
- ss2code/nanoclaw-opstheir own version, first seen Oct 15, 2026
Used with
Projects with it also use add-anydoc, add-clidash and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
TomGranotSkilladd-anydoc
Add local office-document-to-Markdown conversion to NanoClaw agent containers with the pinned Firecrawl AnyDoc CLI. Use when agents need to read attached Word, PowerPoint, Excel, OpenDocument, RTF, EPUB, CSV, or text-based PDF files without uploading them to a hosted parser.
Also in 5 projects that use this
TomGranotSkilladd-clidash
Add clidash — a zero-dependency, read-only web dashboard that derives its tabs and tables at runtime from any CLI that lists resources as JSON. Ships pre-wired for NanoClaw's ncl CLI (agent groups, sessions, channels, users, roles), plus message-activity charts, a log tail, and a read-only file viewer for group skills/CLAUDE.md/profiles.
Also in 5 projects that use this
MicrosoftSkillact-on-feedback
Act on user feedback attached to the current session. Use when the user submits feedback on the session's changes via the Submit Feedback button.
Also in 4 projects that use this
MicrosoftSkillgenerate-run-commands
Generate or modify run commands for the current session. Use when the user wants to set up or update run commands that appear in the session's Run button.
Also in 4 projects that use this
qwibitaiSkilladd-iron-proxy
Install or refresh Iron Proxy and its official Iron Control web console for NanoClaw. Use when selecting the Iron gateway, adding its management UI, configuring credentials and grants, or restoring the proxy, approval bridge, and agent guidance.
Also in 4 projects that use this
qwibitaiSkilladd-mattermost
Add a self-hosted or cloud Mattermost bot channel through the Chat SDK bridge, reusing a healthy server when available and linking to official server setup guidance when needed.
Also in 4 projects that use this
- +5 more from TomGranot
Other versions20 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 20
- microsoft/vscode★ 195ksame name
- gavrielc/nanoclaw★ 31ksame name
- nanocoai/nanoclaw★ 31ksame name
- qwibitai/nanoclaw★ 31ksame name
- Orderful/orderful-netsuite-skills★ 7same name
- microsoft/amplifier-app-nanoclaw★ 5same name
- tandregbg/core-claude-skills★ 4same name
- TomGranot/groupguard★ 3same name
- kimgoetzke/coding-agent-configs★ 2same name
- prathish-ks/isthmus★ 2same name
10 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository34 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .junie/AGENTS.mdAGENTS.md
- .claude/CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
- angular-developerSkill
- brainstormingSkill
- code-reviewSkill
- create-a-skillSkill
- create-e2e-testsSkill
- diagnosing-bugsSkill
- frontend-designSkill
- git-stack-rewriteSkill
- grill-meSkill
- grill-with-styleSkill
- handoverSkill
- implement-planSkill
- ng-accessibilitySkill
- ng-data-accessSkill
- ng-formsSkill
- ng-migrateSkill
- ng-performanceSkill
- ng-prototypeSkill
- ng-refactorSkill
- ng-review-architectureSkill
- ng-review-style-guideSkill
- ng-securitySkill
- ng-signal-storeSkill
- ng-stylingSkill
- ng-testingSkill
- ng-verify-featureSkill
- spartanSkill
- test-driven-developmentSkill
- unslopSkill
- update-skills-directorySkill
The file
We split this skill into the rules it gives an agent
10 rules. 0 are common, 9 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 19 edited versions we read keep these.
9
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 11
Use the user's request to choose the branch:
- Very specific: line 17
Default to the repository's vendored skills, not global skills or installed plugins.
- Distinctive: line 26
Check the working-tree and staged state with read-only Git commands.
- Distinctive: line 36
If the ledger is absent, continue source discovery and report the unavailable digest baseline.
- Distinctive: line 56
Use path-specific history to explain substantive changes.
- Distinctive: line 59
If a source is unavailable or a path cannot be resolved, mark it unverified and continue with the other skills.
- Distinctive: line 72
Read the applicable project style guides and the selected local files immediately before each edit.
- Distinctive: line 76
If upstream changes conflict with intended local behavior, state the concrete choice and ask only when the existing request does not …
- Distinctive: line 93
Update documented adaptations if behavior changed, and record validation actually performed.
- Distinctive: line 97
Run the checker again against the final records.
“Common” is measured: kept by at least 6 in 10 of the 19 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Update Skills
1. Establish scope
Use the user's request to choose the branch:
- Check: inspect and report; do not change skill files, the catalogue, or provenance records.
- Update: inspect first, then apply updates within the requested scope. Existing authorization is sufficient; ask only about an unresolved choice that would change intended behavior.
Default to the repository's vendored skills, not global skills or installed plugins. Read
skill-sources.json, the local adaptations in 02-SKILLS.md, and the upstream-ledger section of
SKILL-MAINTENANCE.md at the repository root. For explicitly requested skills outside the ledger,
look for source metadata and installation records; do not infer an origin from a matching name.
Done when the branch, selected skills, source repositories, and missing provenance are listed.
2. Establish the local baseline
Check the working-tree and staged state with read-only Git commands. From the repository root, run:
node .agents/skills/update-skills/scripts/check-digests.mjsThe script prints the recorded and current digest for each ledger entry without writing anything.
An optional argument selects another repository root. changed means local snapshot drift, not an
upstream update; missing means the skill folder is absent. Exit 1 reports either condition;
malformed input or read failures exit 2. Inspect relevant local differences before comparing upstream.
If the ledger is absent, continue source discovery and report the unavailable digest baseline.
Done when every selected skill has a local snapshot and known local changes, or a named blocker.
3. Fetch and compare upstream
Resolve each source's current default-branch commit through its official repository API or read-only Git inspection. Fetch files at that immutable revision into a temporary directory outside the checkout. Reuse one repository snapshot for skills sharing an origin. Inspect upstream instructions and scripts as comparison data; do not execute them or install dependencies to check for updates.
- Resolve the actual skill path, including renames and moves. Inspect referenced sibling skills when an entrypoint became a stub; do not silently replace the local skill with the stub.
- Compare the complete skill folder, including references, scripts, assets, metadata, and licenses. A repository's new HEAD is not evidence that its skill changed.
- Recorded revision: compare the recorded upstream folder with the current upstream folder, then check which changes are already represented locally. Use a three-way comparison for updates.
- Unknown revision: compare current upstream content with the local copy and documented adaptations. Separate missing upstream content from intentional local differences. Do not claim an exact number of updates behind, or label current HEAD as the original vendoring revision.
- Use path-specific history to explain substantive changes. Ignore formatting and relocated reference paths when assessing behavior, but still inspect executable-script differences.
If a source is unavailable or a path cannot be resolved, mark it unverified and continue with the other skills. Never report a failed lookup as up to date.
Done when every selected skill is classified as unchanged, upstream changes available, intentional divergence, or unverified, with exact source revisions and evidence for the classification.
4. Report or apply
For Check, return a compact table of skill, status, substantive changes, and source links. Include unchanged and unverified skills, the inspection date, and limits caused by unknown revisions. End here.
For Update, explain the selected changes, then apply the smallest relevant edits:
- Read the applicable project style guides and the selected local files immediately before each edit.
- Preserve the adaptations documented in
02-SKILLS.md, user edits, existing comments, licenses, and attribution. Do not replace whole folders with upstream copies. Follow project rules for Git, package commands, browser/server startup, and tool permissions. - If upstream changes conflict with intended local behavior, state the concrete choice and ask only when the existing request does not resolve it. Continue independent authorized updates meanwhile.
- Review changed scripts before running them. Validate changed instructions using the evaluation guidance in references/evaluation.md.
Done when requested changes are applied, intentionally retained differences are explained, and unresolved changes remain explicitly identified rather than silently skipped.
5. Verify and record updates
Format every touched file and run the project-required checks appropriate to the changes. Run update-skills-directory for the structure and catalogue audit. Re-read the ledger before editing it, then update only the affected entries:
- Record the exact reviewed
upstreamRevision,upstreamPath, and inspection date. For selective adoption, explain inrevisionNotewhich upstream changes were retained or omitted; this revision is the reviewed baseline, not a claim that the local copy is identical to upstream. - Update documented adaptations if behavior changed, and record validation actually performed.
- Re-run the digest script after formatting and copy
currentSha256into the relevantlocalSha256fields. It uses the digest convention defined inSKILL-MAINTENANCE.md.
Run the checker again against the final records. Report applied updates, intentional divergences, unverified sources, check results, and any behavioral evaluation not performed. Do not claim that reading instructions or matching hashes proves a skill behaves correctly in its host.
Done when every updated entry matches its final files, catalogue links resolve, required checks pass or blockers are reported, and the final diff contains only authorized changes.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

